********************************************************************************
*Table B2; Column 1 
********************************************************************************
clear all
set matsize 10000
set maxvar 40000
set more off
global data " "
global dir " "
use $data/estimation_sample_blacks, clear



gen dropouts_8th=0  if mothers_education~=.
replace dropouts_8th=1 if mothers_education<=8
keep if dropouts_8th==1



preserve
gen Post0=0
replace Post0=1 if index==0

gen Post1=0
replace Post1=1 if index==1

gen Post2=0
replace Post2=1 if index==2

gen Post3=0
replace Post3=1 if index==3

gen Post4=0
replace Post4=1 if index>=4


global treat_var "Post0 Post1 Post2 Post3 Post4"
global state_var "Exitexam Perstudentexpenditure Studentteacherratio Povertyrate Unemploymentrate"
foreach x in $treat_var{
sum `x', meanonly
scalar mean`x'=r(mean)
}

reg newbw2500 $treat_var $state_var i.age_cat i.tbo i.male  i.division#c.by i.year i.birth_month i.by i.mom_birth_state, cluster(mom_birth_state)

local Z=meanPost0+meanPost1+meanPost2+meanPost3+meanPost4
lincom (_b[Post0]*meanPost0/`Z')+(_b[Post1]*meanPost1/`Z')+(_b[Post2]*meanPost2/`Z')+(_b[Post3]*meanPost3/`Z')+(_b[Post4]*meanPost4/`Z')
restore
********************************************************************************
set seed 23260513
capture postclose placebo		  
postfile placebo b using "$data/table_b2_c1", replace
forvalues i=1/500 {
use $data/estimation_sample_blacks, clear

gen dropouts_8th=0  if mothers_education~=.
replace dropouts_8th=1 if mothers_education<=8
keep if dropouts_8th==1

preserve
bsample, cluster(mom_birth_state)

gen Post0=0
replace Post0=1 if index==0

gen Post1=0
replace Post1=1 if index==1

gen Post2=0
replace Post2=1 if index==2

gen Post3=0
replace Post3=1 if index==3

gen Post4=0
replace Post4=1 if index>=4


global treat_var "Post0 Post1 Post2 Post3 Post4"
global state_var "Exitexam Perstudentexpenditure Studentteacherratio Povertyrate Unemploymentrate"
foreach x in $treat_var{
sum `x', meanonly
scalar mean`x'=r(mean)
}
qui reg newbw2500 $treat_var $state_var i.age_cat  i.tbo i.male  i.division#c.by i.year i.birth_month i.by i.mom_birth_state 
local V=meanPost0+meanPost1+meanPost2+meanPost3+meanPost4
local c_weighted=(_b[Post0]*meanPost0/`V')+(_b[Post1]*meanPost1/`V')+(_b[Post2]*meanPost2/`V')+(_b[Post3]*meanPost3/`V')+(_b[Post4]*meanPost4/`V')
post placebo (`c_weighted')
restore
}

postclose placebo


use table_b2_c1, clear

sum  /****Std Dev is the standard error of the weighted point estimate.*/


********************************************************************************
*Table B2; Column 2 
********************************************************************************
clear all
set matsize 10000
set maxvar 40000
set more off
global data " "
global dir " "
use $data/estimation_sample_blacks, clear



gen dropouts_8th=0  if mothers_education~=.
replace dropouts_8th=1 if mothers_education<=8
keep if dropouts_8th==1



preserve
gen Post0=0
replace Post0=1 if index==0

gen Post1=0
replace Post1=1 if index==1

gen Post2=0
replace Post2=1 if index==2

gen Post3=0
replace Post3=1 if index==3

gen Post4=0
replace Post4=1 if index>=4


global treat_var "Post0 Post1 Post2 Post3 Post4"
global state_var "Exitexam Perstudentexpenditure Studentteacherratio Povertyrate Unemploymentrate"
foreach x in $treat_var{
sum `x', meanonly
scalar mean`x'=r(mean)
}

reg newpretermLT37 $treat_var $state_var i.age_cat i.tbo i.male  i.division#c.by i.year i.birth_month i.by i.mom_birth_state, cluster(mom_birth_state)

local Z=meanPost0+meanPost1+meanPost2+meanPost3+meanPost4
lincom (_b[Post0]*meanPost0/`Z')+(_b[Post1]*meanPost1/`Z')+(_b[Post2]*meanPost2/`Z')+(_b[Post3]*meanPost3/`Z')+(_b[Post4]*meanPost4/`Z')
restore
********************************************************************************
set seed 23260513
capture postclose placebo		  
postfile placebo b using "$data/table_b2_c2", replace
forvalues i=1/500 {
use $data/estimation_sample_blacks, clear

gen dropouts_8th=0  if mothers_education~=.
replace dropouts_8th=1 if mothers_education<=8
keep if dropouts_8th==1

preserve
bsample, cluster(mom_birth_state)

gen Post0=0
replace Post0=1 if index==0

gen Post1=0
replace Post1=1 if index==1

gen Post2=0
replace Post2=1 if index==2

gen Post3=0
replace Post3=1 if index==3

gen Post4=0
replace Post4=1 if index>=4


global treat_var "Post0 Post1 Post2 Post3 Post4"
global state_var "Exitexam Perstudentexpenditure Studentteacherratio Povertyrate Unemploymentrate"
foreach x in $treat_var{
sum `x', meanonly
scalar mean`x'=r(mean)
}
qui reg newpretermLT37 $treat_var $state_var i.age_cat  i.tbo i.male  i.division#c.by i.year i.birth_month i.by i.mom_birth_state 
local V=meanPost0+meanPost1+meanPost2+meanPost3+meanPost4
local c_weighted=(_b[Post0]*meanPost0/`V')+(_b[Post1]*meanPost1/`V')+(_b[Post2]*meanPost2/`V')+(_b[Post3]*meanPost3/`V')+(_b[Post4]*meanPost4/`V')
post placebo (`c_weighted')
restore
}

postclose placebo


use table_b2_c2, clear

sum  /****Std Dev is the standard error of the weighted point estimate.*/

********************************************************************************
*Table B2; Column 3 
********************************************************************************
clear all
set matsize 10000
set maxvar 40000
set more off
global data " "
global dir " "
use $data/estimation_sample_whites, clear



gen dropouts_8th=0  if mothers_education~=.
replace dropouts_8th=1 if mothers_education<=8
keep if dropouts_8th==1



preserve
gen Post0=0
replace Post0=1 if index==0

gen Post1=0
replace Post1=1 if index==1

gen Post2=0
replace Post2=1 if index==2

gen Post3=0
replace Post3=1 if index==3

gen Post4=0
replace Post4=1 if index>=4


global treat_var "Post0 Post1 Post2 Post3 Post4"
global state_var "Exitexam Perstudentexpenditure Studentteacherratio Povertyrate Unemploymentrate"
foreach x in $treat_var{
sum `x', meanonly
scalar mean`x'=r(mean)
}

reg newbw2500 $treat_var $state_var i.age_cat i.tbo i.male  i.division#c.by i.year i.birth_month i.by i.mom_birth_state, cluster(mom_birth_state)

local Z=meanPost0+meanPost1+meanPost2+meanPost3+meanPost4
lincom (_b[Post0]*meanPost0/`Z')+(_b[Post1]*meanPost1/`Z')+(_b[Post2]*meanPost2/`Z')+(_b[Post3]*meanPost3/`Z')+(_b[Post4]*meanPost4/`Z')
restore
********************************************************************************
set seed 23260513
capture postclose placebo		  
postfile placebo b using "$data/table_b2_c3", replace
forvalues i=1/500 {
use $data/estimation_sample_whites, clear

gen dropouts_8th=0  if mothers_education~=.
replace dropouts_8th=1 if mothers_education<=8
keep if dropouts_8th==1

preserve
bsample, cluster(mom_birth_state)

gen Post0=0
replace Post0=1 if index==0

gen Post1=0
replace Post1=1 if index==1

gen Post2=0
replace Post2=1 if index==2

gen Post3=0
replace Post3=1 if index==3

gen Post4=0
replace Post4=1 if index>=4


global treat_var "Post0 Post1 Post2 Post3 Post4"
global state_var "Exitexam Perstudentexpenditure Studentteacherratio Povertyrate Unemploymentrate"
foreach x in $treat_var{
sum `x', meanonly
scalar mean`x'=r(mean)
}
qui reg newbw2500 $treat_var $state_var i.age_cat  i.tbo i.male  i.division#c.by i.year i.birth_month i.by i.mom_birth_state 
local V=meanPost0+meanPost1+meanPost2+meanPost3+meanPost4
local c_weighted=(_b[Post0]*meanPost0/`V')+(_b[Post1]*meanPost1/`V')+(_b[Post2]*meanPost2/`V')+(_b[Post3]*meanPost3/`V')+(_b[Post4]*meanPost4/`V')
post placebo (`c_weighted')
restore
}

postclose placebo


use table_b2_c3, clear

sum  /****Std Dev is the standard error of the weighted point estimate.*/


********************************************************************************
*Table B2; Column 4 
********************************************************************************
clear all
set matsize 10000
set maxvar 40000
set more off
global data " "
global dir " "
use $data/estimation_sample_whites, clear



gen dropouts_8th=0  if mothers_education~=.
replace dropouts_8th=1 if mothers_education<=8
keep if dropouts_8th==1



preserve
gen Post0=0
replace Post0=1 if index==0

gen Post1=0
replace Post1=1 if index==1

gen Post2=0
replace Post2=1 if index==2

gen Post3=0
replace Post3=1 if index==3

gen Post4=0
replace Post4=1 if index>=4


global treat_var "Post0 Post1 Post2 Post3 Post4"
global state_var "Exitexam Perstudentexpenditure Studentteacherratio Povertyrate Unemploymentrate"
foreach x in $treat_var{
sum `x', meanonly
scalar mean`x'=r(mean)
}

reg newpretermLT37 $treat_var $state_var i.age_cat i.tbo i.male  i.division#c.by i.year i.birth_month i.by i.mom_birth_state, cluster(mom_birth_state)

local Z=meanPost0+meanPost1+meanPost2+meanPost3+meanPost4
lincom (_b[Post0]*meanPost0/`Z')+(_b[Post1]*meanPost1/`Z')+(_b[Post2]*meanPost2/`Z')+(_b[Post3]*meanPost3/`Z')+(_b[Post4]*meanPost4/`Z')
restore
********************************************************************************
set seed 23260513
capture postclose placebo		  
postfile placebo b using "$data/table_b2_c4", replace
forvalues i=1/500 {
use $data/estimation_sample_whites, clear

gen dropouts_8th=0  if mothers_education~=.
replace dropouts_8th=1 if mothers_education<=8
keep if dropouts_8th==1

preserve
bsample, cluster(mom_birth_state)

gen Post0=0
replace Post0=1 if index==0

gen Post1=0
replace Post1=1 if index==1

gen Post2=0
replace Post2=1 if index==2

gen Post3=0
replace Post3=1 if index==3

gen Post4=0
replace Post4=1 if index>=4


global treat_var "Post0 Post1 Post2 Post3 Post4"
global state_var "Exitexam Perstudentexpenditure Studentteacherratio Povertyrate Unemploymentrate"
foreach x in $treat_var{
sum `x', meanonly
scalar mean`x'=r(mean)
}
qui reg newpretermLT37 $treat_var $state_var i.age_cat  i.tbo i.male  i.division#c.by i.year i.birth_month i.by i.mom_birth_state 
local V=meanPost0+meanPost1+meanPost2+meanPost3+meanPost4
local c_weighted=(_b[Post0]*meanPost0/`V')+(_b[Post1]*meanPost1/`V')+(_b[Post2]*meanPost2/`V')+(_b[Post3]*meanPost3/`V')+(_b[Post4]*meanPost4/`V')
post placebo (`c_weighted')
restore
}

postclose placebo


use table_b2_c4, clear

sum  /****Std Dev is the standard error of the weighted point estimate.*/

